R

RWKV7 Goose Pile 168M HF

Developed by RWKV
RWKV-7 model using Flash Linear Attention format, trained on the Pile dataset, supporting English text generation tasks.
Downloads 57
Release Time : 3/17/2025

Model Overview

This is a 168M-parameter RWKV-7 model using Flash Linear Attention format, primarily designed for English text generation tasks.

Model Features

Flash Linear Attention Format
Utilizes Flash Linear Attention format to enhance model efficiency.
Dynamic State Evolution
Supports expressive dynamic state evolution to improve model performance.
Efficient Training
Employs bfloat16 format and cosine decay learning rate for efficient training.

Model Capabilities

Text Generation
Language Modeling

Use Cases

Text Generation
Open-Domain Text Generation
Generates coherent and meaningful English text.
Language Understanding
Language Model Evaluation
Evaluates language understanding capabilities on benchmarks like LAMBADA and PIQA.
LAMBADA accuracy 45.6%, PIQA accuracy 65.5%
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase